[R-meta] Questions on Implementing CHE and SCE Models in R

[Fée Ona Fuchs MSH Hamburg]

Hi everyone,

thank you for the opportunity to ask for help here!
I am conducting a meta-analysis examining the association between Childhood Maltreatment (CM) and Emotion Recognition (ER). In order to prevent distortions by the conversion between effect size metrics, I analyse Group differences (Hedges’ g) between individuals with high and low CM and Associations (Pearson's r) between CM severity and ER separately.

For both, I also run separate models for:
(a) Accuracy rates, and
(b) Reaction times measured via ER tasks

In many cases, there are multiple effect sizes from a single study. These are calculated for:
- Different emotions assessed (e.g., anger, sadness, happiness).
- Different CM subtypes (e.g., emotional abuse, physical neglect).

Research Questions
I’m using here the example of analysis 1a (group differences in accuracy rates).
I want to investigate the following:
(a) Overall group differences: Are there general ER differences between adults with low and high CM (across all emotions)?
(b) Emotion-specific differences: Do group differences vary between specific emotions?
(c) CM subtype-specific differences: Do group differences vary between specific CM subtypes?
(d) Moderator effects: Do variables such as % female participants, age, clinical sample (0,1), or stimulus material (0,1) influence the overall effect?

Example dataset structure (analysis 1a) (n=18, k=74, here: 6 rows):
(av=emotions; uv=CM-Subtype)

structure(list(
  study = c("Study 1", "Study 1", "Study 2", "Study 2", "Study 2", "Study 2"),
  av = structure(c(10L, 10L, 4L, 1L, 5L, 9L), levels = c("1", "2", "3", "4", "5", "6", "7", "9", "10", "11", "12", "13", "14"), class = "factor"),
  uv = structure(c(5L, 6L, 1L, 1L, 1L, 1L), levels = c("1", "2", "3", "4", "5", "6", "7", "8", "9", "10", "11"), class = "factor"),
  n = c(75, 75, 38, 38, 38, 38),
  female = c(32, 32, NA, NA, NA, NA),
  age = c(NA, NA, NA, NA, NA, NA),
  SP_clinic = c(1, 1, 0, 0, 0, 0),
  stimuli = c(1, 1, 1, 1, 1, 1)
), row.names = c(NA, -6L), class = c("tbl_df", "tbl", "data.frame"))

(a) General Group Differences in ER (Overall Effect)

# Calculate SE^2
df_g_acc$vi <- df_g_acc$hedges_g_se^2

# Constant sampling correlation assumption
rho <- 0.6

# Impute covariance matrix
V_g_acc <- with(df_g_acc,
          impute_covariance_matrix(vi = vi,
                                   cluster = study,
                                   r = rho))

# Fit model
che.g_acc <- rma.mv(effectsize_hedges_g ~ 1,
                    V = V_g_acc,
                    random = ~ 1 | study/no,
                    data = df_g_acc,
                    sparse = TRUE)

# Confidence Intervals
conf_int(che.g_acc,
         vcov = "CR2")
coef_test(che.g_acc,
          vcov = "CR2")

# I²
i2_che_g_acc <- var.comp(che.g_acc)
i2_che_g_acc

# Robust F-test
Wald_g_acc <- Wald_test(che.g_acc,
                        constraints = constrain_zero(1),
                        vcov = "CR2")
Wald_g_acc


(b) Emotion-Specific Group Differences (SCE)

# Impute covariance matrix for subgroup emotions
V_g_acc_emo <- impute_covariance_matrix(df_g_acc_filtered$vi,
                                        cluster = df_g_acc_filtered$study,
                                        r = rho,
                                        smooth_vi = TRUE,
                                        subgroup = df_g_acc_filtered$av)

# Fit random effects working model
sce_g_acc_emo <- rma.mv(effectsize_hedges_g ~ 0 + av,
                      V = V_g_acc_emo,
                      random = list(~ av | study), struct = "DIAG",
                      data = df_g_acc_filtered, sparse = TRUE)
sce_g_acc_emo

# Confidence Intervals
CI_g_acc_emo <- conf_int(sce_g_acc_emo, vcov = "CR2")
CI_g_acc_emo

# Robust F-test
Wald_g_acc_emo <- Wald_test(sce_g_acc_emo,
                            constraints = constrain_equal(1:7),
                            vcov = "CR2")
Wald_g_acc_emo

(c) Subtype-Specific Group Differences (SCE)
For this question, I performed a similar SCE, replacing subgroup = df_g_acc_filtered$av with subgroup = df_g_acc_filtered$uv.


Questions regarding the CHE model:
- Heterogeneity: Did I calculate the I^2 statistics correctly using the var.comp() function?
- Moderators: To answer (d), would you recommend extending the CHE model (a) by adding the moderators (e.g., % female, age, clinical sample, and stimuli material) as additional predictors? Or would you recommend calculating a separate SCE to answer this question?

SCE:
- Does the way I have specified the model seem appropriate?
- For the correct interpretation of the results, I just want to make sure I’m on the right track: To determine whether group differences differ significantly between specific emotions, is it correct to rely on the confidence intervals (CIs) excluding zero and the Wald test results? Apologies if this is a basic question—I just want to make sure I’m interpreting the findings properly.
I’ve noticed that, while I have significant effects for single emotions in the SCE, the Wald test indicates no significance, suggesting there is no difference between the emotions. Could you help clarify why this might be the case? Is it common for these results to differ, and how should I interpret these discrepancies in the context of my model?

Thank you so much for your time and input! I’d greatly appreciate any feedback, especially on whether I’ve implemented the SCE and covariance matrix correctly.

Kind regards,
Fée

__________________________________________________________________________

Fée O. Fuchs
PhD student, Clinical Psychologist (M.Sc.)

MSH Medical School Hamburg
University of Applied Sciences and Medical University
Am Kaiserkai 1
20457 Hamburg

www.medicalschool-hamburg.de
feeona.fuchs using medicalschool-hamburg.de
-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/pkcs7-signature
Size: 5508 bytes
Desc: not available
URL: <https://stat.ethz.ch/pipermail/r-sig-meta-analysis/attachments/20241216/bd0942b3/attachment.p7s>